A modified hybrid DCT-SVD image-coding algorithm

Today's image compression technique is experiencing many improvements under the method of transform coding. Many research groups have developed the image-coding algorithms that optimize bit rate and image quality. The objective of this thesis is to propose the hybrid image-coding algorithm that combines discrete cosine transform (DCT) with singular value decomposition (SVD). A decision criterion as to the choice of DCT or SVD is standard deviation (STD) of the 8 x 8 blocks of the image. DCT is used to transform the high correlation or low standard deviation 8 x 8 blocks. SVD is used to transform the low correlation or high standard deviation 8 x 8 blocks. Adaptive multistage vector quantization (AMVQ) is used to encode the eigenvectors of the 8 x 8 blocks of the image using SVD. Uniform quantization is used to encode the eigenvalues of the 8 x 8 blocks using SVD and the DCT coefficients of the 8 x 8 blocks. The simulation results show good image quality at low bit rates.